Intelligent and Robust UAV-Aided Multiuser RIS Communication Technique With Jittering UAV and Imperfect Hardware Constraints

• Published in: IEEE transactions on vehicular technology 2023-08, Vol.72 (8), p.10737-10753
• Main Authors: Adam, Abuzar B. M., Wan, Xiaoyu, Elhassan, Mohammed A. M., Muthanna, Mohammed Saleh Ali, Muthanna, Ammar, Kumar, Neeraj, Guizani, Mohsen
• Format: Article
• Language: English
• Subjects: active beamforming; Array signal processing; Artificial neural networks; Autonomous aerial vehicles; Beamforming; Communication; Communication networks; convolutional long short-term memory (CLSTM); Deep learning; Hardware; Iterative methods; neural network; Neural networks; Optimization; Power consumption; reconfigurable intelligent surface (RIS); Taylor series; Trajectory; UAV trajectory; Unmanned aerial vehicle (UAV); Unmanned aerial vehicles
• ISSN: 0018-9545; 1939-9359
• DOI: 10.1109/TVT.2023.3255309

In this paper, we investigate unmanned aerial vehicle (UAV)-aided multiuser reconfigurable intelligent surface (RIS) communication for next generation communication networks. We aim to jointly optimize the active beamforming, passive beamforming, and UAV trajectory jointly to minimize power consumption in presence of UAV jitters and imperfect hardware constraints. We decouple the formulated nonconvex problem into three subproblems. For active beamforming subproblem, we linearize and approximate the constraints using S-procedure and general sign-definitiveness technique. Then, we again apply S-procedure and convex-concave technique to handle the passive beamforming. For UAV trajectory subproblem, we apply first-Taylor expansion to transform the problem into a tractable form. On the highlights of the proposed solution, we design a hybrid semi-unfolding deep neural network (HSUDNN) to mitigate the constraints during the channel state information gain for RIS and UAV links in real-time. Using our proposed active beamforming solution and the optimality conditions, we design the unfolding-based sub neural network. Moreover, we design inception-like multi-kernel convolutional long short-term memory (IL-MK-CLSTM) sub networks to handle the UAV trajectory and passive beamforming. IL-MK-CLSTM provides spatiotemporal connection which helps in overcoming vanishing gradient problem and provides multi-step prediction. The proposed HSUDNN achieves 99.24% accuracy which demonstrates its superior performance in comparison to the existing state-of-the-art techniques in literature.